buildingopen

Bouncer — Security skill for Claude Code

Security community intermediate

Independent quality gate that uses Gemini to audit Claude Code's output.

How to install Bouncer

This entry records only its repository, not the path inside it, so there is no exact command to give. Open buildingopen/bouncer and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Bouncer does

Gemini stands at the door. Independent quality gate that audits Claude Code's output before it can stop. Score below threshold? Claude keeps working.

Alternatives in Security

Security Guidance — Hook that warns about command injection, XSS, and unsafe patterns 14k ★
Security Guardian — Comprehensive security analysis for RTK CLI tool, focusing on command injection, shell escaping, hook security 11.9k ★
Review Artifact — Creates one polished, self-contained HTML artifact from review findings, audit notes, PR feedback, code review 652 ★

README

Bouncer

Gemini stands at the door. Independent quality gate that audits Claude Code's output before it can stop. Score below threshold? Claude keeps working.

[](https://github.com/buildingopen/bouncer/actions/workflows/test.yml) [](https://buildingopen.github.io/bouncer/) [](https://github.com/buildingopen/bouncer/releases) [](https://github.com/buildingopen/bouncer/blob/master/LICENSE)

Why it gets shared

Claude cannot wave through unverified work. Another model checks the evidence.
One install gives you both the automatic Stop hook and the on-demand /bouncer skill.
The repo is explicit about security, data flow, and how to test the installer in a clean environment.

Live site

https://buildingopen.github.io/bouncer/

User prompt → Claude Code → [Stop Hook] → Gemini 2.5 Flash
                                              ↓
                                         Score 1-10
                                              ↓
                                    ┌─────────┴─────────┐
                                    │                    │
                               Score = 10           Score < 10
                                    │                    │
                              ✓ Approve              ✗ Block
                           (Claude stops)      (Claude keeps working
                                                with Gemini's feedback)

What it looks like

Quick audit:

========================================
  BOUNCER AUDIT: 9/10
========================================

SCORE: 9/10
ISSUES:
- missing explicit test command output in final message
VERDICT: FAIL

Deep audit:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    BOUNCER DEEP AUDIT  [###########################...]  9/10
    GUEST LIST: almost flawless
    Verified in 12.4s
  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~